Planning for object storage deployment

There are a number of decisions that must be made before beginning your object storage deployment in an IBM Spectrum Scale™ environment. The following is an outline of these decision points.

The IBM Spectrum Scale for object storage architecture is shown in Figure 1.

Each protocol node runs all OpenStack Swift object services, the Keystone identity service, and the IBM Spectrum Scale client. Client applications make requests to perform object operations such as uploading an object, downloading an object, and deleting an object. The request is routed to a protocol node typically by a load balancer or by using DNS round robin. The protocol node implements that request by creating, retrieving, or deleting the object on the backend IBM Spectrum Scale storage. The request completion status is then returned to the client application. Each client request must include an authentication token. Typically, client applications first request a token from the Keystone service, and then provide that token in the subsequent object requests, until the token expires. At that point, a new token can be requested.

For more information on OpenStack Swift, see the OpenStack Swift documentation.

For more information on OpenStack Keystone as it is used with Swift, see the Keystone Auth section in the OpenStack Swift documentation.